
PLIN0020 Advanced Semantic Theory Yasu Sudo Classifier Languages Autumn 2018 1 Classifier Languages: Basics There’s a class of languages called (obligatory) classifier languages, e.g. Mandarin Chi- nese, Vietnamese, Japanese, Korean, etc. Not every classifier language is a East Asian language, e.g. Yucatec Mayan. See World Atlas of Language Structures for more informa- tion. Obligatory classifier languages are so called because numerals obligatorily appear with classifiers when they modify nouns (but see §?? for some complications). (1) Mandarin Chinese (Sino-Tibetan) yì běn shū * yì shū sān běn shū one cl book one book three cl book ‘one book’ ‘three books’ yì zhī māo * yì māo sān zhī māo one cl cat one cat three cl cat ‘one cat’ ‘three cats’ (2) Japanese (Japonic) hana ichi-rin * hana ichi hana san-rin flower one-cl flower one flower three-cl ‘one flower’ ‘three flowers’ kuruma ichi-dai * kuruma ichi kuruma san-dai car one-cl car one car three-cl ‘one car’ ‘three cars’ (3) Vietnamese (Austroasiatic) một con chó * một chó hai con chó one cl dog one dog two cl dog ‘one dog’ ‘two dogs’ Mandarin, Cantonese, and Japanese are believed to have hundreds of classifiers (although not all of them are frequently used). Wikipedia has lists for Mandarin and Cantonese and Japanese. I don’t know how many classifiers other classifier languages have. Many classifiers have selectional restrictions that seem to be semantic in nature. For ex- ample, the classifier -wa in Japanese goes with any noun describing birds (or rabbits). (4) Japanese karasu ichi-wa niwatori ichi-wa pengin ichi-wa crow one-cl chicken one-cl penguin one-cl ‘one crow’ ‘one chicken’ ‘one penguin’ This is an example of taxonomic restriction. Sometimes the selectional restriction is based on the function (e.g. -dai in Japanese for vehicles of various kinds such as cars, bikes, etc.), 1 and sometimes on the shape (-mai in Japanese for flat objects) of objects being described. Some classifiers refer to multiple aspects, e.g. taxonomy and shape: e.g. -hiki for small animals (that are not birds), -too for big animals in Japanese. (5) Japanese neko go-hiki tokage go-hiki kuzira go-too cat five-cl lizard one-cl whale five-cl ‘five cats’ ‘five lizards’ ‘five whales’ Some nouns are compatible with multiple classifiers. They often have semantic differ- ences: (6) Japanese inu yon-hiki inu yon-too dog five-cl dog five-cl ‘five small dogs’ ‘five big dogs’ Both Mandarin and Japanese have ‘general purpose classifiers’: • Mandarin ge can be used basically with any noun • Japanese -tsu can be used with many inanimate noun. When highly frequent classifiers are applicable, a use of the general purpose classifier is often judged infelicitous. There’s also a class of ‘container classifiers’ (alt.: ‘massifiers’) that denote particular ways of packaging things, and are similar in function to container nouns like bottle and glass in English. These are called classifiers too, because at least morphologically they seem to pattern with other classifiers. (7) Japanese hon san-hako abura san-teki juusu san-kan book three-cl.box oil three-cl.drop juice three-cl.can ‘three boxes of books’ ‘three drops of oil’ ‘three cans of juice’ 1.1 Sanches-Greenberg-Slobin Generalization Classifier languages generally lack obligatory number marking (see the data above). This generalization is called the Sanches-Greenberg-Slobin Generalization, after Greenberg (1972) and Sanches & Slobin (1973)(Doetjes 2012 mentions some controversial counter- examples). • The SGS generalization is about obligatory number marking. Optional number mark- ing is allowed, in classifier languages and in fact attested, e.g. Chinese men, Japanese reduplication. • It’s a one-way implication: every classifier language lacks obligatory number marking, but not every language that lacks obligatory number marking is a classifier language. 2 1.2 No number-marking, but no classifiers Dëne Su̧łiné (a.k.a. Chipewyan; spoken in Canada) has no obligatory number marking (Wilhelm 2008). (8) Larry ʔı̧łághe ʔejëre nághénígh. Larry ʔejëre nádághéłnígh. Larry one bovine perf.buy.O Larry bovine distr.perf.buy.O ‘Larry bought one cow.’ ‘Larry bought several cows.’ (Wilhelm 2008:45) But it also lacks classifiers: (9) so̧lághe k’ásba so̧lághe bek’eshích’elyı̧ five chicken five table ‘five chickens’ ‘five tables’ (Wilhelm 2008:46) There are other languages like this, e.g. Tagalog, Ojibwe and Yudja (a.k.a. Juruna; spoken in Xingu, Brazil) (Doetjes 2012, Mathieu 2012, Lima 2014). One difference between Dëne Su̧łiné and Tagalog vs. Ojibwe and Yudja is that in the latter languages any noun can be directly modified by numerals. (10) Dëne Su̧łiné * so̧lághe ʔejëretth’úé * so̧lághe bër five milk five meat (Wilhelm 2008:46) (11) Yudja a. Txabïu pïza dju wï. Three canoe bring ‘(Someone) brought three canoes.’ b. Txabïu apeta dju wï. Three blood bring ‘(Someone) brought three quantities of blood.’ (Lima 2014:10) 1.3 Optional classifier languages Western Armenian (Indo-European) is a famous example of optional classifier languages (Borer 2005, Khanjian 2009, Bale, Gagnon & Khanjian 2011, Bale & Barner 2012). (12) Yergu (had) hovanoc uni-m. two (cl) umbrella have-1sg ‘I have two umbrellas.’ (Borer 2005:94) Interestingly, when a classifier is present, the noun must be singular. (13) a. Yergu hovanoc-ner uni-m. two umbrella-pl have-1sg ‘I have two umbrellas.’ 3 b. *Yergu had hovanoc-ner uni-m. two cl umbrella-pl have-1sg (Borer 2005:94) However, this requirement that nouns be singular with numerals might be independent from optional classifiers. It is observed in languages like Turkish and Hungarian as well. 2 Linguistic Relativism Benjamin Whorf and and John A. Lucy, in particular, claim that nouns in classifier lan- guages uniformly describe unindividuated and uncountable entities, and that speakers of these languages perceive the world differently from speakers of languages like English. In particular, see John A. Lucy’s work on Yucatec, an obligatory classifier language (Lucy 1992a,b, 1997, Lucy & Gaskins 2001, Lucy 2004). A strong version of this view (linguistic determinism; Whorf 1956, Quine 1960) holds that speakers of classifier languages (and other languages with no mass-count distinction) lit- erally don’t classify things into discrete entities vs. non-discrete substances. But this does not seem to be tenable: • Obviously, speakers of classifier languages can count (with or without linguistic expres- sions). • There are grammatical phenomena that require counting, e.g. more. • The distinction between discrete vs. non-discrete entities seems to be fundamental to cognition. Pre-linguistic infants make this distinction and appears to employ it in word learning (Baillargeon, Spelke & Wasserman 1985, Soja, Carey & Spelke 1991, Wynn 1992, Spelke, Kestenbaum, Simons & Wein 1995, Xu & Carey 1996, Feigenson, De- haene & Spelke 2004, Carey 2009). • Some lexical items encode this and related ontological distinctions in some ways, e.g. many languages distinguish eat vs. drink. A weaker version would be to say that whether you are a speaker of a classifier language or not has certain specific effects on your perception. Whether this is the case and if so what the effects actually are is actively researched (Imai & Gentner 1997, Lucy & Gaskins 2001, Imai & Mazuka 2007, Barner, Inagaki & Li 2009, Li, Dunham & Carey 2009). Here are some major findings. Imai & Gentner’s (1997) paradigm (cf. Soja et al. 1991, Lucy 1992b): You are presented with a reference item with a nonse noun. Your task is to pick one of two objects that you think is describable by the same noun. One of the two objects has the same shape but is made of a different material, the other object has a different shape, but is made of the same material. (14) Look at this blicket. Point to the tray that also has the blicket on it. 4 They tested Japanese and American children and adults. Four age categories: Japanese American Early 2-yo (mean 2;1) 1;10–2;5 (n=14) 2;1–2;5 (n=14) Late 2-yo (mean 2;8) 2;7–3;2 (n=15) 2;6–3;0 (n=15) 4-yo (mean 4;2) 3;9–4;7 (n=14) 3;10–4;6 (n=14) Adults U. of Kyoto students (n=18) Northwestern students (n=18) • Speakers of both languages, including 2 year-olds, distinguish concrete/discrete enti- ties vs. substances. • Japanese speakers gave material-based answers more often, especially in the Simple Objects and Substances conditions. One might wonder if this has something to do with learning new words. Maybe English speakers tend to treat new nouns as count nouns for completely independent reasons, e.g. there are more count nouns in the English lexicon? To this end, Imai & Mazuka (2003, 2007) conducted variants of the above experiment involving no nouns. 5 (15) Look at this. Which is the same as this? Imai & Mazuka (2007) observe similar language effects in this task: However, some argue that the deictic pronoun this still requires speakers to postulate a hypothetical noun so the results might be just a matter of word learning and lexical statistics (Barner et al. 2009, Li et al. 2009). Li et al. (2009) conducted several similar experiments with Mandarin speakers, with or without nonse words, and observe similar effects of language (‘*’ means ‘statistically signif- icant’; Experiment 2 has more types of solid objects classified according to [˘C(omplex)] and [˘F(unctional)]): 6 However, when they explicitly asked whether the subjects see the same objects as Exper- iment 2 as a solid entity or a substance, the language effects disappeared. (16) When we see something, we tend to think of it as either an object or a substance. [EXAMPLES] We’re interested in the factors that lead someone to think of something as either an object or a substance [...].
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages14 Page
-
File Size-